Sex determines which section of the SLC6A4 gene is linked to obsessive e compulsive symptoms in normal Chinese college students

Previous case e control and family-based association studies have implicated the SLC6A4 gene in obsessive e compulsive disorder (OCD). Little research, however, has examined this gene ’ s role in obsessive e compulsive symptoms (OCS) in community samples. The present study genotyped seven tag SNPs and two common functional tandem repeat polymorphisms ( 5-HTTLPR and STin2 ), which together cover the whole SLC6A4 gene, and investigated their associations with OCS in normal Chinese college students ( N ¼ 572). The results revealed a signi ﬁ cant gender main effect and gender-speci ﬁ c genetic effects of the SLC6A4 gene on OCS. Males scored signi ﬁ cantly higher on total OCS and its three dimen- sions than did females ( ps < .01). The 5-HTTLPR in the promoter region showed a female-speci ﬁ c genetic effect, with the l/l and l/s genotypes linked to higher OCS scores than the s/s genotype ( ps < .05). In contrast, a conserved haplotype polymorphism ( rs1042173 j rs4325622 j rs3794808 j rs140701 j rs4583306 j rs2020942 ) covering from intron 3 to the 3 0 UTR of the SLC6A4 gene showed male-speci ﬁ c genetic effects, with the CGAAGG/CGAAGG genotype associated with lower OCS scores than the other genotypes ( ps < .05). These effects remained signi ﬁ cant after controlling for OCS-related factors including partici- pants ’ depressive and anxiety symptoms as well as stressful life events, and correction for multiple tests. These results are discussed in terms of their implications for our understanding of the sex-speci ﬁ c role of the different sections of the SLC6A4 gene in OCD.


Background
Obsessiveecompulsive disorder (OCD) is a psychiatric disorder characterized by clinically significant recurrent, intrusive and disturbing thoughts and images and/or repetitive ritualistic physical or mental acts (APA, 2007;Mathews et al., 2004). Approximately 2e3% of people in the world are suffering from OCD (Kessler et al., 2005a(Kessler et al., , 2005b. As a prevalent psychiatric disorder, OCD has a median onset age of around 20 years old (Brynska and Wolanczyk, 2005;Mathews et al., 2004), which suggests that college students between 18 and 25 years old are at higher risk than other age groups to develop OCD.
OCD patients usually display one or more common clusters of obsessiveecompulsive symptoms (OCS), such as cleanliness and contamination, obsessions, compulsions, symmetry, and ordering (Bloch et al., 2008a;Mathews et al., 2004;Moore et al., 2010). Moreover, continuous distributions of these OCS in the general population exist with few obsessiveecompulsive behaviors and minimal severity at one end and many obsessiveecompulsive behaviors and severe impairment at the other (Apter et al., 1996;Brynska and Wolanczyk, 2005).
OCD has a substantial genetic basis (Walitza et al., 2010), which implies that OCS in normal population would also likely have some genetic influence. A better understanding of the genetic basis of OCS in normal populations, especially in the highly susceptible college students, could provide insights into the pathology of OCD and thereby, promote more effective methods of prevention, diagnosis, and treatment of OCD. The estimated heritability for OCD ranges from 0.45 to 0.65 for children and adolescents and from 0.27 to 0.47 for adults in twin and family studies (van Grootheest et al., 2005), indicating that early-onset OCD may have a different etiology (including genetic factors) from late-onset OCD. Molecular genetic association studies have specifically provided further evidence for the genetic contributions of SLC6A4 to OCD (Bloch et al., 2008b;Hu et al., 2006;Lin, 2007;Voyiaziakis et al., 2011). SLC6A4 transports serotonin from the synaptic cleft to presynaptic neurons, thereby attenuating current serotonin neurotransmission while also maintaining the pool of available serotonin for subsequent release (Lesch and Gutknecht, 2005). SLC6A4 is also the molecular target of selective serotonin reuptake inhibitors (SSRIs), the most clinically effective medications for OCD treatment (APA, 2007;Geller et al., 2003;Goddard et al., 2008). Brain imaging studies have also suggested that late-onset OCD, but not the earlyonset OCD, is associated with abnormally low serotonin transporter availability in the brain (Hasselbalch et al., 2007;Hesse et al., 2011). Therefore, the SLC6A4 gene is an ideal candidate gene for research on OCD/OCS. When examining OCS gender is an important variable, not only because of gender differences in OCD/OCS but also because of potential interactions between gender and genes. Growing literature shows that sex differences are prevalent in the neural and genetic bases of behaviors (Cahill, 2006). Such differences have been linked to not only sex chromosomes (X-and Y-linked genes and proteins) and gonadal hormone secretions (Ngun et al., 2011), but also autosomal chromosomes and general neurochemical mechanisms (Bocklandt and Vilain, 2007;Sanchez and Vilain, 2010). For example, recent studies found that gender modulated the associations of catechol-o-methyltransferase (COMT) genetic variants with brain function (Harrison and Tunbridge, 2008) and personality traits (Chen et al., 2011), and the associations of monoamine oxidase A (MAOA) genetic variants with brain circuitry underling aggression and personality traits Buckholtz and Meyer-Lindenberg, 2008). More specifically, gender Â gene interactions involving the SLC6A4 gene have been frequently reported for behaviors such as depression (Baune et al., 2008;Brummett et al., 2008) and anxiety (Mizuno et al., 2006). Of most relevance to our study, three previous association studies directly reported gender-specific genetic effects of SLC6A4 variants on OCS (Dickel et al., 2007;Voyiaziakis et al., 2011) and obsessiveecompulsive personality disorder (OCPD) (Blom et al., 2011), although the specific interaction effects were not consistent across studies (a point to which we will return in the discussion). Therefore, the current study specifically examined gender and gender-by-gene (SLC6A4) interaction effects on OCS.
The SLC6A4 gene has two well-known common functional polymorphisms: one in the upstream regulatory region (5-HTTLPR) and the other (the 17 bp tandem repeat, STin2) in intron 2 (Fiskerstrand et al., 1999;Heils et al., 1996). The s allele of 5-HTTLPR has been linked to reduced SLC6A4 expression and low serotonin reuptake compared to the l allele (Heils et al., 1996). The 10R allele of STin2 has been associated with lower transcriptional activity than the 12R allele (Fiskerstrand et al., 1999). Earlier studies Hu et al., 2006;Kim et al., 2005;McDougle et al., 1998) and a recent review (Bloch et al., 2008b) reported that the l allele of 5-HTTLPR was a risk factor for OCD while some other studies reported that the s allele was a risk factor (Lin, 2007;Perez et al., 2006). As mentioned early, interactions of SLCA64 by gender on OCD (Dickel et al., 2007;Voyiaziakis et al., 2011) and OCPD (Blom et al., 2011), were also reported. Two studies reported no association (Camarena et al., 2001;Chabane et al., 2004).
There are three limitations in previous studies. First, they were either caseecontrol or family-based studies, with clinical OCD samples of heterogeneous age of onset and medication treatments (e.g., SSRIs). These patients were also comorbid with various psychiatric disorders, such as major depression, panic disorder and anxiety disorder (Perez et al., 2006;Voyiaziakis et al., 2011), which would not allow for a separation of the SLC6A4 effects on OCS from related psychiatric symptoms. Only one recent association study examined 5-HTTLPR and obsessiveecompulsive personality disorder (OCPD) trait in a community sample of European ancestry, with the finding that the s allele was linked to lower obsessiveecompulsive personality trait scores in males and to higher obsessiveecompulsive trait scores in females (Blom et al., 2011). The second limitation of previous research is that most studies were done with European ancestry populations (Bloch et al., 2008b;Blom et al., 2011;Lin, 2007;Voyiaziakis et al., 2011). Considering that allele frequencies of SLC6A4 polymorphisms have substantial differences between Asian and European ancestry individuals (Gelernter et al., 1997;Zhong et al., 2009), these allele frequency diversities may reflect different behavioral adaptive functions in different cultures and living environments (Chiao and Blizinsky, 2010) or differential physiological functions of the SLC6A4 gene in different populations (Smits et al., 2004). The third limitation of previous research is that much attention was focused on 5-HTTLPR without examining polymorphisms in other sections of the SLC6A4 gene. Only a few studies have included STin2 and several other single nucleotide polymorphisms (Voyiaziakis et al., 2011;Wendland et al., 2008), and none included normal community samples.
The current study was designed to investigate the role of the SLC6A4 gene in OCS in normal Chinese college students. To cover the whole gene, we included the often-studied 5-HTTLPR, STin2, and seven tag single nucleotide polymorphisms (SNPs) (see Fig. 1 and Table 1). Finally, to untangle the role of the SLC6A4 gene in OCS versus related affective disorder symptoms such as depression and anxiety, we further analyzed our results by taking into account these symptoms as well as a major environmental factordstressful life events (Carter et al., 2004;Middeldorp et al., 2007;Ogilvie et al., 1996;Quarantini et al., 2011).

Participants
Five hundred and seventy-two (312 females) undergraduate students (all Han Chinese, 20.47 AE 1.01 years old) were recruited from Beijing Normal University (Beijing, China) (He et al., 2010). All participants had normal or corrected-to-normal vision, no neurological and psychiatric problems based on self-reports. Informed written consent was obtained from each participant. The current study was approved by the Beijing Normal University Institutional Review Board.

Behavioral measures
Participants were asked to complete the 20-item Leyton Obsessional Inventory-Child Version (LOI-CV) on a 4-point scale of symptom frequency (always ¼ 3, mostly ¼ 2, sometimes ¼ 1, and never ¼ 0) (Berg et al., 1988), which has been widely used to measure obsessiveecompulsive symptoms in epidemiological studies (Maggini et al., 2001;Mather and Cartwright-Hatton, 2004;Roussos et al., 2003). The Chinese translation of LOI-CV was done by a team of several bilingual psychologists and had high internal consistency (Cronbach a ¼ 0.829). In addition, participants were also asked to completed the 21-item Beck Depression and Anxiety Inventory (BDI and BAI) (Beck, 1990;Beck et al., 1996), and a scale of stressful life events (Beam et al., 2002;Wills et al., 1992). The Cronbach a's were 0.809 for BDI, 0.89 for BAI, and 0.746 for stressful life events.

Genotyping
A 4 ml venous blood sample was collected from each participant. Genomic DNA was extracted according to the standard method within 2 weeks. 5-HTTLPR and STin2 were genotyped by using the standard polymerase chain reaction technique (PCR). Detailed PCR protocols for 5-HTTLPR (He et al., 2010) and STin2 (Li et al., 2007) were described in previous studies. Finally, PCR products of both 5-HTTLPR (the l allele of 528 bp and the s allele of 484 bp), and STin2 (10R of 267 bp and 12R of 300 bp) were separated by 2% agarose gels stained with ethidium bromide and then analyzed with a Gel Doc 2000 imaging system (Bio-Rad Laboratories Ltd, UK).
The seven tag SNPs of the SLC6A4 gene (see Table 1 and Fig. 1) were genotyped using the standard Illumina GoldenGate Genotyping protocol (www.southgene.com.cn) in 480 participants selected randomly out of the total sample. They all passed the criteria of a call rate of >95%, Minor Allele Frequency (MAF) of >0.05, and HardyeWeinberger equilibrium of ps > .05.
The OCS score was normally distributed in the current population (Skewness ¼ 0.54, Kurtosis ¼ 0.48) and all participants were located within Mean AE 3SD. An exploratory factor analysis of the current data showed similar three-factor structure as reported in previous studies: Compulsions, Obsessions, and Cleanliness (Bamber et al., 2002;Moore et al., 2010). Thus, in the current study, three dimension scores were also calculated for each participant. Age range of the current sample was narrow (18e26 years) and showed no significant effect on OCS, thus was excluded in subsequent analysis.  As shown in Fig. 1, six SNPs (rs1042173, rs4325622, rs3794808, rs140701, rs4583306 and rs2020942) were in strong linkage and constructed a conserved haplotype block (D 0 > 0.95) covering from intron 3 to 3 0 UTR of the SLC6A4 gene. Three main haplotypes, CGAAGG, AAGGAG and AAGGAA (rs1042173j rs4325622j rs3794808j rs140701j rs4583306j rs2020942) showed respective frequencies of 0.78, 0.12 and 0.08. The homozygotes of the CGAAGG haplotype (n ¼ 301) were assigned the label CGAAGG/CGAAGG, whereas the homozygotes and heterozygotes of the minor and other rare haplotypes were merged to a combined non-CGAAGG/CGAAGG group. 1 In terms of 5-HTTLPR, individuals with l/l (n ¼ 52) and l/s (n ¼ 219) genotypes were merged together as in previous research (Kim et al., 2005). For Stin2, there were few participants with 10R/ 10R (n ¼ 3), 12R/13R (n ¼ 2), 12R/14R (n ¼ 2) and 12R/15R (n ¼ 2). Thus, 10R/10R (n ¼ 3) and 10R/12R (n ¼ 68) were merged together as the 10R carriers while 12R/12R (n ¼ 458) and others were merged as the non-10R carriers.
General linear models (GLM) in SPSS for Windows (version 16) were used to examine the main and interactive effects of genetic variations and gender on OCS. To further investigate whether potential covariates could have accounted for such effects, we added depression, anxiety, stressful lift events into the above models as covariates. Finally, in order to examine whether genetic effects were specific for particular dimensions or all dimensions of OCS, we repeated the above analyses for each of the three dimensions of OCS.
Due to the linkage disequilibrium in the SLC6A4 gene, we used the Bonferroni correction procedure proposed by Nyholt (2004). The effective number (Meff) of independent marker loci in the three genetic variants (the haplotype, STin2 and 5-HTTLPR) was calculated using the SNPSpD method (Nyholt, 2004), yielding an Meff of 2.49. Thus, for the general linear models of OCS, the significant threshold was adjusted by dividing the conventional alpha 0.05 by 2.49, resulting in 0.02. For the general linear models of three dimensions of OCS, the significant threshold was adjusted by dividing the conventional alpha 0.05 by the effective number of tests (N ¼ 3 Â 2.49), resulting in 0.007.

Results
Significant gender main effects were found in OCS scores, ps < .01. Males scored higher than females in total OCS score as well as three dimensions. As shown in Table 2 and the left panel of Fig. 2, different polymorphisms of the SLC6A4 gene were associated with OCS for the two genders. 5-HTTLPR showed female-specific genetic effects on OCS, with the female minor allele carriers (l/l and l/s) displaying higher scores on total OCS than the female major allele homozygotes (s/s), p ¼ .008. This result remained significant after taking into account OCS-related factors including depression, anxiety, and stressful life events, p ¼ .004. Furthermore, these results survived correction for multiple tests. For males, however, 5-HTTLPR was not significantly associated with OCS. This gender specificity was supported by a significant interaction between 5-HTTLPR and gender, F ¼ 7.171, p < .01. These results were consistent for the dimensions of Cleanliness (F ¼ 4.450, p < .05) and Obsessions (F ¼ 6.49, p < .05).
In contrast to the female-specific effect of 5-HTTLPR, the haplotype block (rs1042173j rs4325622j rs3794808j rs140701j rs4583306j rs2020942) and STin2 showed male-specific genetic effects (see Supplementary Table S1 for the same pattern of data for each SNP). As shown in Table 2 and the right panel of Fig. 2, male CGAAGG/CGAAGG carriers showed lower scores in total OCS (marginally at p ¼ .085 without covariates and significantly at p ¼ .024 with covariates of depression, anxiety, and stressful life events), and especially in the Cleanliness dimension (p ¼ .004 without covariates and .002 with covariates), than did the male non-CGAAGG/CGAAGG carriers. The male-specific genetic effect of the haplotype on Cleanliness dimension also remained significant after correction for multiple tests. The interaction between the haplotype and gender on Cleanliness was marginally significant, F ¼ 2.782, p < .10. Male 10R carriers of STin2 scored significantly higher in Cleanliness dimension than non-10R carriers, p ¼ .04 without covariates and p ¼ .014 with covariates. However, the male-specific effect of the STin2 on Cleanliness did not survive  correction of multiple tests. There were no significant associations between the SLC6A4 genetic variants and depression, anxiety and stress life events.

Discussion
The current study was the first to investigate gender-specific effects of the whole SLC6A4 gene on OCS in a large sample of normal Chinese college students. It was also the first to consider the confounding effects of depression, anxiety and stressful life events when examining genetic effects on OCS. We found significant gender differences in OCS, with males scoring higher in total OCS and its three dimensions than did females. These results were consistent with previous studies that used various scales measuring for OCS and OCD (Richter et al., 1994). Interestingly, both the current study and that of Lensi et al. (1996) found that gender differences were especially large in the dimension of Obsessions. It should be noted that age of onset for OCD has been found to differ by gender, with males having an earlier onset. For example, a recent study of a Chinese clinical OCD sample found that male patients showed an average onset age of around 20 years while females around 27 years (Li et al., 2009). Considering that the current sample had an average age of 20 years, the gender differences in OCS we found may be specific to this age group.
Second, the current study found that different sections of the SLC6A4 gene were related to OCS for males and females. The promoter region's polymorphism (5-HTTLPR) was related to females' OCS, but the haplotype block containing the gene (and to some extent, STin2) was related to males' OCS. This intriguing finding is in line with the growing literature of gender-specific genetic effects on behaviors and neural correlates of behaviors (Cahill, 2006). Biological and physiological differences between the sexes in sex chromosomes (X-and Y-linked genes and proteins), gonadal hormone secretions (Ngun et al., 2011), and autosomal chromosomes and general neurochemical mechanisms (Bocklandt and Vilain, 2007;Sanchez and Vilain, 2010) can all contribute to gender differences in behaviors. Recent studies have documented specific examples of gender-modulated genetic effects of catecholo-methyltransferase (COMT) on personality traits (Chen et al., 2011), and those of the monoamine oxidase A (MAOA) on brain circuitry underlying aggression (Buckholtz and Meyer-Lindenberg 2008) and harm avoidance . For the SLC6A4 gene, researchers have found gender-specific associations of 5-HTTLPR with psychiatric disorders such as depression (Baune et al., 2008;Brummett et al., 2008) and anxiety (Mizuno et al., 2006). A clinical pharmacological study also found genderspecific effects of 5-HTTLPR on treatment effectiveness of selective serotonin reuptake inhibitors (SSRIs) on depression (Smits et al., 2008), suggesting intrinsic gender-specific biochemical and physiological differences in the function of 5-HTTLPR. In terms of OCS/ OCD, our results on 5-HTTLPR were in line with a recent caseecontrol review study in which the l allele of 5-HTTLPR was shown to be significantly associated with OCD in females (Dickel et al., 2007). The link between the l allele and OCD is consistent with the hypothesis that abnormally low serotonin neurotransmission in the brain plays an important role in the pathology of OCD since the l allele has been reported to be associated with high SLC6A4 expression, consequently leading to a relatively low serotonin neurotransmission Heils et al., 1996;Hranilovic et al., 2004). Clinical and pharmacological studies found that selective serotonin reuptake inhibitors (SSRIs), which facilitate serotonin neurotransmission, are effective choices for medical treatments of OCD (Geller et al., 2003;Goddard et al., 2008). OCD patients also showed abnormally reduced serotonin transporter availability in limbic and paralimbic brain areas, especially in the late-onset OCD patients (Hasselbalch et al., 2007;Hesse et al., 2011). Not all studies, however, found this pattern of results. For example, the only other study conducted with a community sample found that the OCPD trait was linked to the s allele among females, but the l allele among males (Blom et al., 2011). A number of differences between this study and the current study, including ethnicity, age range of the two samples, and different measures used, may have contributed to this inconsistency. As mentioned earlier, substantial diversities in the allele frequencies of 5-HTTLPR in different ethnic populations may indicate different behavioral adaptive functions of the SLC6A4 polymorphisms in different cultures and different living environments (Chiao and Blizinsky, 2010). Much more research is needed to understand the dynamic interactions between genes and environments.
In contrast to 5-HTTLPR, both STin2 and the haplotype block (rs1042173j rs4325622j rs3794808j rs140701j rs4583306j rs2020942) showed male-specific genetic effects on OCS. Male 10R carriers (mostly 10R/12R) of STin2 showed higher scores on Cleanliness dimension than did the non-10R males. Male non-CGAAGG/CGAAGG carriers showed higher scores on total OCS and Cleanliness than did male CGAAGG/CGAAGG carriers. Few previous studies have examined STin2 and none studied the haplotype in terms of their potential associations with OCS or OCD. A recent study found that 9R/10R was significantly associated with OCD in European ancestry females (Voyiaziakis et al., 2011), consistent with our results albeit for a different gender. Two other caseecontrol studies, one with Spanish (Baca-Garcia et al., 2007) and the other with Japanese (Ohara et al., 1999) showed that the 12R allele was associated with the clinical patient group consisting of OCD, panic, phobias and general anxiety disorders in the former study and 12R/12R and 10R/ 12R were associated with OCD as compared to other psychiatric patients and controls in the latter study. However, these two studies did not analyze data by gender.
As for the specific evolutionary, biochemical or physiological mechanisms of the gender Â haplotype interactions on OCS, we can only make some reasonable speculations. From the HapMap data (www.hapmap.org; also see Supplementary Figure S1), the current haplotype block belongs to a huge highly-linked region spanning part of the SLC6A4 gene, the whole NSrp70 gene (also termed the CCDC55gene) (Kim et al., 2011), and the EFCAB5 gene. This large linkage disequilibrium (LD) block suggested that it has experienced strong recent selection (Wang et al., 2006). The haplotype linked to lower OCS showed stronger LD than the haplotype linked to higher OCS, suggesting positive selection for low OCS. Although evolutionary psychologists (Abed and de Pauw, 1998) have speculated that moderate versions of compulsive behavior (e.g., frequent checks of hygiene and the presence of enemies, hoarding) might have had evolutionary advantages during early periods of human evolution, our study suggests recent deselection of compulsive behaviors. It should be noted that, however, that because this is a large LD block, the selection of a particular haplotype at this locus may have acted on any of the three genes in the block. Future research can address this intriguing conjecture of recent deselection of OCS/OCD and the role of other genes.
Finally, because the SLC6A4 gene has been linked to several affective disorders, it was not clear whether its associations with OCS/OCD were due to the latter's shared variance with other disorders. In this study, we reanalyzed our data by controlling for several common confounding factors such as depression, anxiety, and stressful life events, and showed that the SLC6A4 gene's link may be specific to OCS/OCD, particularly its Cleanliness dimension. Future studies should take into consideration the different subtypes of OCS/OCD. Three main limitations of the current study need to be mentioned. First, the current study did not genotype tri-allelic 5-HTTLPR. A single nucleotide polymorphisms rs25331 on the l allele results in three alleles: l A , l G and s (Hu et al., 2006;Wendland et al., 2008), in which l G and s both reduce SLC6A4 expression. Nevertheless, the frequency of the minor G allele of rs25531 in Asians was estimated to be 9.1% (Zhang et al., 2009). Thus, although the current study did not differentiate l G and l A (s/s: n ¼ 298, l/s: n ¼ 219; l/l ¼ 52), the low frequencies of the l allele and the G allele of rs25531 were not expected to dramatically alter the results. Second, this was a gene-behavior association study in normal college students. The biochemical and physiological functions of the haplotype block (rs1042173j rs4325622j rs3794808j rs140701j rs4583306j rs2020942) and the linked functional loci were not explored in the present study. Previous studies found that variations in intron or noncoding regions can also exert transcriptional suppression of the genes, influence mRNA secondary structure (Nackley et al., 2006), and regulate gene expression and splicing . Such biochemical work needs to be conducted specifically for the haplotype identified in the current study. Third, the present study was conducted with one ethnic group (i.e., Han Chinese college students) to avoid potential population stratification problem. These results should be compared to the results of other ethnic populations in the future. Finally, we limited our sample to a homogeneous group of college students. Psychologists have lamented for decades on the "college sophomore" phenomenon in social psychological research (Henry, 2008;Sears, 1986).
Because of potential geneeenvironment interactions, it is not clear whether our results can be generalized to other age groups. Nevertheless, college enrollment (or selection effects) may have more to do with characteristics such as intelligence, motivation, and ideologies than with OCS, because OCS covers many areas of daily life that are common inside and outside of the college environment. Furthermore, college enrollment is gradually becoming the norm rather than the exception even in China. That said, future behavioral and genetic studies of OCS should directly compare college and non-college students and provide information about the role of environments in OCS and in modulating the effects of genes on OCS.
In conclusion, the current study observed the gender-specific effects of the SLC6A4 polymorphisms on OCS in normal Chinese college students. These effects were independent of depression, anxiety and stressful life events. The haplotype block (rs1042173j rs4325622j rs3794808j rs140701j rs4583306j rs2020942) spanning from intron 3 to 3 0 UTR and STin2 in intron 2 showed male-specific genetic effects on OCS, with the non-CGAAGG/CGAAGG genotypes associated with higher OCS scores. Male 10R carriers of STin2 had higher Cleanliness scores compared to male non-10R carriers. In contrast, 5-HTTLPR showed female-specific genetic effects, with the l/l and l/s genotypes linked to higher OCS scores. These findings provided specific evidence for the gender-specific role of the SLC6A4 gene in the continuum of OCS, confirmed the contributions of disturbed serotonin neurotransmission to the pathology of OCD, and should be of relevance to the diagnosis and treatment of OCD.

Role of the funding source
This study was supported by the 111 Project of the Ministry of Education of China (B07008).

Contributors
Chuansheng Chen, Robert K. Moyzis, Qi Dong, Jun Li, Gui Xue designed the study. Qinghua He, Xuemei Lei, Jin Li, He Li, Chunhui Chen, Bi Zhu, Zhongyu Cao, Xiongying Chen, wrote the protocol and collected genetic and behavioral data. Sufang Li took part in collecting of genetic data. Xuemei Lei undertook the analysis and managed the literature searches and wrote the first draft of the manuscript. Anna Shan Chun Hsu took part in editing the manuscript. All authors contributed to have approved the final manuscript.